A structured knowledge base that gives AI agents methodology for every stage of product development. Four frameworks, sixteen skills, one traceable artifact chain.
Install all 16 skills into any project
Copies skills to .agent/skills/ · patches CLAUDE.md · safe to re-run
Every stage produces a traceable artifact. Every artifact links forward and back. Nothing falls through the cracks.
OKRs, OST, discovery
PRD → USM → USL → USD → UAT
WF-XXX → COMP-XXX
Next.js 15 + ShadCN
Quality gates on all artifacts
Each framework is a complete knowledge base — rules, templates, and examples. AI agents load them automatically via skills.
Six PM disciplines with embedded methodology from Teresa Torres, Marty Cagan, and Strategyzer. From vision to go-to-market.
Five-stage pipeline from product brief to BDD test cases. Every story gets acceptance criteria. Every AC gets a test case. Fully auto-traced artifact chain.
Text-based UI design. 4 themes. WF-XXX wireframes and COMP-XXX specs without leaving Markdown.
Next.js 15 App Router patterns. Build working frontends with dummy data before investing in backend.
Each skill is a markdown file that activates when you use a trigger keyword. No configuration needed. Works with Claude Code, Codex, Gemini, and Cursor.
pm-product-strategy
pm-product-discovery
pm-market-research
pm-data-analytics
pm-marketing-growth
pm-go-to-market
po-brief-to-prd
po-prd-to-usm
po-usm-to-usl
po-usl-to-usd
po-usd-to-uat
validate-prd
validate-usd
design-wireframe
design-component-spec
Skills are agent-agnostic markdown. Each AI agent gets its own entry point into the same knowledge base.
Auto-loads via .agent/skills/
Skill index via AGENTS.md
Skill index via GEMINI.md
Rules file via .cursorrules
All agents read from skills/*/SKILL.md — the single agent-agnostic source of truth. One knowledge base, every agent.
Desired Outcome
↳ Increase activation rate by 20%
Opportunity 1
↳ Users don't understand value at signup
→ Solution A: Interactive onboarding tour
→ Solution B: Video demo on welcome screen
Opportunity 2
↳ Setup takes too long (> 10 min)
→ Solution C: One-click template import
Riskiest assumption:
Users want templates (Evidence: None 🔴)
Grounded in Teresa Torres' Continuous Discovery, Marty Cagan's product thinking, and Strategyzer's business model tools. Run a discovery sprint before a single story is written.
Map user needs → solutions → riskiest assumptions. Don't jump to features.
Surface riskiest assumptions. Generate JTBD interview scripts to test them.
Strategy Canvas, OKR hierarchy, value proposition, launch plan, battlecards.
Every story gets acceptance criteria. Every AC gets a test case. The `/po-pipeline` command runs all five stages automatically, pausing for your review at the right moments.
Problem, goals, personas, success metrics.
User activities and steps, visually mapped.
Prioritized backlog with MoSCoW method.
Atomic, observable, binary acceptance criteria per story.
BDD scenarios (Given / When / Then).
---
artifact: USD
generated-by: po-usl-to-usd
---
## UI Elements
- AC-001: Search input visible with placeholder
- AC-002: Results list shows title + excerpt
## UI Behavior
- AC-003: Results update on keystroke (300ms debounce)
## Non-Functional
- NFR-001: Response < 200ms for < 10k records
Install all 16 skills into your project in one command. Works with Claude Code, Codex, Gemini, and Cursor.